I. Introduction ........................................................ 119

II. Facts and Procedural History ....................................... 120
    A. Introduction .................................................... 120
    B. Initial Taxation of Blue Cross Blue Shield Entities ............. 120
    C. Capital's Contracts in 1987 ..................................... 121
       1. Premium structures ........................................... 121
       2. Renewals ..................................................... 122
       3. Valuation .................................................... 122
    D. The Present Controversy ......................................... 122

III. Deductibility of Losses on Health Insurance Contracts ............. 124
    A. The Blue Cross Blue Shield Fresh Start Basis Rule ............... 124
    B. Deductibility of Non-Sale Losses by Blue Cross Blue Shield
         Organizations ................................................. 125
    C. Are the Insurance Contracts Mass Assets? ........................ 125

IV. Burden of Proof of Deduction Amounts ............................... 128
    A. Who Bears the Burden of Proof? .................................. 128
    B. Proof of Separate Valuation ..................................... 129

V. Capital's Valuation ................................................. 131
    A. The Reinsurance Model ............................................ 131
       1. Highest and best use .......................................... 131
       2. Separate values .............................................. 132
       3. The goodwill adjustment ...................................... 133
    B. Specific Contract Characteristics ............................... 134
    C. Lapse Rates and "Lifing Analysis" ............................... 136
       1. Prospective changes in the market ............................ 137
       2. Human factors ................................................ 139
       3. The Commissioner's lifing arguments .......................... 139
    D. The Commissioner's Objections ................................... 140
    E. Summary and Conclusions ......................................... 140

Capital Blue Cross ("Capital") appeals from a decision of the United States Tax Court denying its request for a refund of overpayment of taxes for tax year 1994. Capital claims that it properly established a basis in hundreds of insurance contracts that were terminated in that year, and that it is therefore entitled to take a loss deduction under 26 U.S.C. § 165 to account for the cancellation of those contracts. The Tax Court found that Capital had not established its basis in those contracts; it therefore treated Capital's basis as zero and denied any deduction.

We agree with Capital that the Tax Court improperly discounted expert testimony that tended to establish Capital's basis in the disputed contracts, and that the zero basis found by the Court was inconsistent with the facts and hence clearly erroneous. Capital engaged in an extensive and professional valuation process in order to calculate its basis in the lost contracts. While the Tax Court correctly found that the Commissioner of Internal Revenue ("the Commissioner") established several flaws in Capital's valuation, overall, we are convinced that Capital's process was thorough and professional, and that it arrived at an essentially reasonable valuation for the cancelled contracts. Given these conclusions, we are unwilling to affirm the Tax Court merely because we find some flaws in Capital's valuation process. Instead, we will reverse and remand for further proceedings.

We leave it to the Tax Court to find the correct valuation for Capital's contracts. On remand, the Commissioner may again press his objections to Capital's methods, and the Tax Court may consider those objections in arriving at a final valuation. But the existence of some problems in Capital's valuation process will not justify finding a zero basis in the lost contracts. Instead, the Tax Court must do its best to calculate a reasonable and correct basis; the Commissioner can best assist the Court by raising specific and quantifiable objections to Capital's valuation, and by proposing alternative methods that will lead to what, in his submission, would be a more reasonable valuation. Thus far, the Commissioner has pointed to alleged flaws in the valuation methodology without explaining or quantifying how they impacted the bottom-line calculation, and without offering any alternatives. We conclude that, on the facts before us, such a procedure is insufficient to reject Capital's claimed deductions.

As suggested above, this case concerns the procedures under which Blue Cross Blue Shield organizations may take loss deductions for terminated subscriber contracts. Since Blue Cross Blue Shield organizations became taxable in 1986, this issue has slowly grown in importance. It has only recently reached the attention of the courts and the Internal Revenue Service ("IRS" or "the Service"). The Service has

inform[ed] Blue Cross Blue Shield insurance organizations that the Service will challenge deductions for losses that relate to the termination of individual customer, provider, or employee contracts or relationships associated with customer lists, provider networks, and workforce in place with respect to which the taxpayer claims an adjusted basis derived from section 1012(c)(3)(A)(iii) [sic] of the Tax Reform Act of 1986.

I.R.S. Notice 2000-34, 2000-2 C.B. 172.

We are the second Article III court to consider the deductibility of these losses. The first case was Trigon Insurance Co. v. United States, 215 F. Supp. 2d 687 (E.D.Va.2002), which found for the Commissioner and disallowed the deductions. The court ultimately held that the taxpayer — Trigon, a success or to two Blue Cross Blue Shield insurers — had not established the 1987 fair market value of its insurance and provider contracts and was therefore not entitled to a deduction. The decision has not been appealed, and our analysis will be guided in part by Judge Payne's thorough opinion. Since Trigon, the IRS has reaffirmed and clarified the position of Notice 2000-34 in a Coordinated Issue Paper dated May 27, 2005. See 2005 WL 1412148 (I.R.S.).

Capital Blue Cross is a Blue Cross Blue Shield organization that sells health insurance to individuals and groups in central and northeastern Pennsylvania. It was founded in 1938, became a Blue Cross organization in 1972, and became a Blue Cross Blue Shield licensee when Blue Cross and Blue Shield merged in 1982.

Historically, Blue Cross Blue Shield organizations were not subject to federal income taxes. In 1986, the Congress, concerned that this gave Blue Cross Blue Shield organizations an unfair competitive advantage over other for-profit insurers, eliminated the tax exemption effective January 1, 1987. See Tax Reform Act of 1986, Pub.L. No. 99-514, § 1012, 100 Stat.2085, 2390-94 (codified at 26 U.S.C. §§ 501(m) & 833). When Blue Cross Blue Shield entities became taxable, they needed to have a way to determine their tax basis in their assets. Congress therefore provided that Blue Cross Blue Shield organizations could take a basis step-up in their assets, so that their tax basis in each asset would be its fair market value ("FMV") on January 1, 1987. Tax Reform Act of 1986, § 1012(c)(3)(A)(ii). This basis step-up is sometimes called the "Fresh Start Basis Rule," see Trigon, 215 F.Supp.2d at 691.

On January 1, 1987, when Capital became a taxable entity, it had 23,526 group health insurance contracts outstanding, as well as a number of individual contracts. Many groups had entered into multiple contracts with Capital; the 23,526 group contracts represented a total of 12,579 separate insured groups.

Under each of these group health insurance contracts, Capital agreed to provide health insurance coverage to the individual members of each group (typically the employees of a company), and the group agreed to pay premiums. Individual group members could elect the type of insurance benefit (individual, single parent with dependents, or family) and coverage (basic medical, basic hospital, major medical, or comprehensive) that they wanted to receive.

Each group was charged a total annual premium; the group decided what portion of the premium was to be paid by the employer and what was to be paid by the individual group members. The premium for a group was determined in one of three ways.

Groups with fewer than one hundred individual members were community-rated. Under community rating, premiums for a particular type of coverage and benefit were based on the cumulative claims history of all of Capital's community-rated contracts for that coverage/benefit. Claims experience for all community-rated contracts in one year would serve as the basis for the premiums on community-rated contracts in the following year. Some 90 percent of Capital's group contracts were community-rated.

Groups of over one hundred members had their premiums determined based on their own claims experience, rather than on the shared claims experience of multiple contracts. Thus, premiums for each such group would be unique. Capital had two ways of setting these premiums.

In experience-rated contracts, total claims received from the members of a group in one year would form the basis for the premiums charged to that group in the following year. If an experience-rated group paid premiums in one year that were excessive relative to its claims, it could receive a "retrospective refund" (cash) or a "retrospective credit" (credited against the next year's premiums). Experience-rated contracts made up over 60 percent of the total value of Capital's contracts.

Cost-plus-rated contracts were a variation on experience-rated contracts; premiums for the following year were calculated based on claims in the previous year plus Capital's administrative costs related to the group. Cost-plus contracts also had a retrospective adjustment feature that would adjust premiums in a given year to more closely match claims and administrative costs in that year.

The experience-rated and cost-plus rated contracts are, for our purposes, basically identical. For convenience, we will sometimes refer to the experience-rated and cost-plus-rated contracts together as "separately-rated" contracts, as distinguished from community-rated contracts whose premiums are determined collectively.

Capital's community-rated contracts were automatically renewed (unless cancelled) on a month-to-month basis; its experience-rated and cost-plus contracts were automatically renewed on an annual basis. The Tax Court found that groups could cancel their contracts at will: the customer group could just stop paying the premiums, which would cause Capital to cancel the contract. Capital's CEO testified, however, that few contracts were cancelled for that reason, and opined that Capital might have a cause of action against a customer who simply stopped paying premiums while still under a contract.

Because premiums were adjusted every year, Capital could typically expect to make money on most of its contracts, since it was not locked into premium rates that might prove inadequate. But Capital was not guaranteed a profit. A separately-rated contract could build up a retrospective deficit (i.e., claims could be higher than premiums in a given year). Premiums would then be increased, but if claims continued to increase faster than premiums, the deficit could constantly increase. Because groups could essentially cancel their contracts at will, and because Capital seems to have had no claim against the group for any accumulated claims-based deficit that was not paid, Capital could in fact lose money on any contract. The Tax Court cited one example: the Pennsylvania Farmer's Union had experience-rated, retrospective contracts with Capital that had a cumulative deficit of $700,000 in January 1988. By 1994, the deficit had reached $4 million. Anxious to recoup this loss, Capital proposed a 48% rate increase for the following year. Pennsylvania Farmer's Union rejected the rate increase and terminated its contracts, leaving Capital to bear the $4 million loss.

From 1938 through 1986, when Capital was a tax-exempt entity, it had never valued its group insurance contracts for tax purposes. It also did not compute a cost basis for the contracts in its financial records, as the contracts were self-created assets. When it became taxable in 1987, Capital did not make any adjustments in its tax books and records to reflect the basis step-up allowed by the 1986 tax changes. In fact, it did not make any such adjustments until 1994; prior to that, its records did not reflect any basis in the group contracts, and it claimed no § 165 deductions for any losses on those contracts until that year.

As of January 1, 1987, Capital was the leading health insurance provider in its geographical area. Alternatives to Blue Cross Blue Shield — such as health maintenance organizations (HMOs), preferred provider organizations (PPOs), and other health insurance products — were not nearly as widespread or important as they are today. But they were developing in importance, and by 1987 Capital was facing increased competition from HMOs and PPOs. Capital's management was aware of this increasing competition, and the company's weakening future prospects, in 1987.

As noted above, Capital did not take any steps to apply the fresh start basis rule when it came into effect in 1987: its tax books did not reflect the stepped-up basis of its insurance contracts.1 This changed in September 1995, when Capital filed its 1994 tax return. This return claimed $2,648,249 of loss deductions for subscriber groups who cancelled their insurance contracts with Capital. The $2.6 million loss was based on a 1995 valuation by Deloitte & Touche. The losses came from the 1994 cancellation of 376 of the over 12,000 group contracts that Capital had owned in 1987. At roughly the same time, Capital filed amended tax returns for 1991-1993, claiming loss deductions for contracts cancelled during those years.

The Commissioner disallowed the deductions in a notice of deficiency dated August 16, 2001. The notice stated a number of theories under which the Commissioner rejected the deductions, including (1) that "it has not been established that any abandonment occurred during the taxable year, or that any loss was sustained"; (2) that the customer contracts were "components of intangible assets which constitute single indivisible assets"; and (3) that the basis step-up applies only to sale or exchange losses, not to abandonment losses.

On November 13, 2001, Capital filed a petition in the United States Tax Court challenging the Commissioner's deficiency determination. The petition claimed contract loss deductions for 1994 totaling $3,342,944. As the Tax Court correctly noted, Capital did not explain why it now claimed this amount, rather than the some $2.65 million claimed on its tax return. In preparation for trial, Capital retained a valuation expert, Daniel McCarthy, who is a member of the actuarial consulting firm Milliman USA and a highly credentialed actuary.2 McCarthy calculated that Capital's basis in the contracts cancelled in 1994 was $3,973,022.94. McCarthy also calculated values for contracts cancelled in other years. Capital now claims deductions totaling over $37 million for tax years 1991-2000, including the roughly $4 million for the 1994 tax year. While this case directly concerns Capital's $4 million claimed deduction for 1994, it will affect at least $37 million in deductions for Capital.

A trial was held in March and April of 2003 before Judge Stephen J. Swift. Expert testimony was taken from McCarthy and other experts for Capital and for the Internal Revenue Service (hereinafter the "Service" or "IRS"). After trial, Judge Swift entered a decision for the Commissioner, dated March 12, 2004. See Capital Blue Cross v. Comm'r, 122 T.C. 224, 2004 WL 473263 (2004). The court rejected the third theory in the Commissioner's notice of deficiency (that the basis step-up does not apply to § 165 losses), was equivocal on the second (that the contracts were part of "indivisible assets"), but agreed with the Commissioner that Capital had not established any loss, and therefore upheld the disallowance of the entire deduction. Capital timely appealed.

The Tax Court had jurisdiction over Capital's petition under 26 U.S.C. §§ 6213, 6214, 6512, and 7442. We have appellate jurisdiction under 26 U.S.C. § 7482(a)(1). We have plenary review over the Tax Court's legal conclusions, and may set aside findings of fact if they are clearly erroneous. Neonatology Associates, P.A. v. Comm'r, 299 F.3d 221, 227 (3d Cir.2002).

III. Deductibility of Losses on Health Insurance Contracts

Loss deductions are governed by § 165 of the Internal Revenue Code, which provides in relevant part:

(a) General rule. — There shall be allowed as a deduction any loss sustained during the taxable year and not compensated for by insurance or otherwise.

(b) Amount of deduction. — For purposes of subsection (a), the basis for determining the amount of the deduction for any loss shall be the adjusted basis provided in section 1011 for determining the loss from the sale or other disposition of the property.

26 U.S.C. (hereinafter "I.R.C.") § 165(a)-(b).

The main dispute in this case concerns the value of Capital's lost contracts, and who bears the burden of proving that value. But we begin with more fundamental questions. First, is the cancellation of valuable contracts a "loss" under § 165(a), such that Capital is entitled to a deduction? Second, if so, what was Capital's basis in those contracts under § 165(b)? It is useful to begin with the second question.

Normally, a taxpayer's basis in an asset is the taxpayer's cost of acquiring that asset. I.R.C. § 1012; see also id. § 1011. A taxpayer's basis in a self-created customer asset — e.g., an insurer's basis in an insurance contract with its customers — will normally be zero. The taxpayer may generally deduct the costs of acquiring the asset (advertising, underwriting, etc.) when those costs are incurred, but does not capitalize them into a basis in the asset. On the other hand, an insurance company might have a cost basis in insurance contracts that it purchased from another insurer. Its basis in such contracts would be the purchase price.

With respect to contracts written after 1987, Capital follows these principles. Thus, the parties have stipulated that "for tax accounting purposes beginning in 1987, Petitioner expensed/deducted the cost of securing customer insurance contracts in the year such expenses were incurred." Capital would thus have zero basis in self-created contracts acquired after 1987: it deducts the costs of acquiring those contracts, and so has no cost basis.

However, with respect to the assets that it possessed on January 1, 1987, Capital's basis is not a cost basis but the fair market value of those assets on that date. The basis step-up provision for Blue Cross Blue Shield organizations states that, "for purposes of determining gain or loss, the adjusted basis of any asset held on the 1st day of [the taxpayer's first taxable year beginning after December 31, 1986] shall be treated as equal to its fair market value as of such day." Tax Reform Act of 1986, § 1012(c)(3)(A), Pub.L. No. 99-514, 100 Stat.2085, 2394. Capital's tax year is the calendar year, meaning that its basis in its pre-1987 assets should be determined based on their fair market value on January 1, 1987.

The purpose of the basis step-up provision was, as the Tax Court noted, "to prevent Blue Cross Blue Shield organizations from being taxed on appreciation in the value of assets that had occurred in pre-1987 years when the organizations had not been subject to Federal income tax." 122 T.C. 234 (citing H. Conf. Rep. 99-841, at II-350 (1986), 1986-3 C.B. (vol.4) 1, 350). A Blue Cross Blue Shield organization that bought a building for $10,000 in 1980, saw its value increase to $50,000 in 1987, and sold it for $65,000 in 1990, would have taxable income of only $15,000. A typical taxpayer in the same situation would have $55,000 in taxable income on the sale, but Congress did not want to tax the Blue Cross Blue Shield organizations for unrealized appreciation that occurred before they became taxable.

This basic understanding of Capital's basis in its assets raises several further questions. To that end, in Part II.B, we inquire whether the basis step-up applies to § 165 losses, and in Part II.C, we discuss its application to intangible assets like the insurance contracts at issue here.

Congress's intent in passing the basis step-up seems primarily to have been to provide tax relief for gains on sales. But basis is also used to calculate deductible losses when a taxpayer sells an asset for less than its basis. The basis found under §§ 1011 and 1012 is also used in calculating deductions when the asset is "lost" under § 165. Losses of assets used in business are deductible under that section, as detailed in the regulations:

A loss incurred in a business or in a transaction entered into for profit and arising from the sudden termination of the usefulness in such business or transaction of any nondepreciable property, in a case where such business or transaction is discontinued or where such property is permanently discarded from use therein, shall be allowed as a deduction under section 165(a) for the taxable year in which the loss is actually sustained.

Treas. Reg. § 1.165-2(a).

In the Tax Court, and in the notice of deficiency, the Commissioner reasoned that the Blue Cross Blue Shield basis step-up applied only to sales transactions, and not to § 165 loss situations like the one presented here. See 122 T.C. 234-35. He cited to the legislative history of the Tax Reform Act of 1986, which includes the statement that "The basis step-up is provided solely for the purposes of determining gain or loss upon sale or exchange of the assets, not for purposes of determining amounts of depreciation or for other purposes." Id. at 235 (quoting H. Conf. Rep. 99-841, at II-349 to 350, 1986-3 C.B. (vol.4) 1, 349-50).

The Tax Court rejected this contention. Judge Swift found that the statutory language unambiguously provides a basis step-up for the purposes of determining any gain or loss, and therefore declined to look to the legislative history to change the meaning of an unambiguous statute. Id. at 236. He also noted that the policy rationale behind the step-up applies in cases of § 165 losses. See id. at 237. The Commissioner has not appealed this conclusion, and we accept it. We note that it is in accord with the only other reported decision on the question, Trigon, 215 F.Supp.2d at 699-701.

Before we turn to Capital's disputed valuation, we address another issue that was raised in the Tax Court but not pursued with clarity on appeal. That is the question whether the contracts at issue here were actually assets subject to § 165 losses at all. While the Commissioner does not clearly raise this argument here, its assumptions underlie many of his objections to Capital's valuation.

It seems to be agreed that these contracts are in fact assets, in that each contract constitutes the right to a continuing stream of future payments. See Trigon, 215 F.Supp.2d at 696; cf. Newark Morning Ledger Co. v. United States, 507 U.S. 546, 113 S. Ct. 1670, 123 L. Ed. 2d 288 (1993) (holding that a newspaper subscription list was a depreciable asset); Union Bankers Ins. Co. v. Comm'r, 64 T.C. 807, 1975 WL 3176 (1975), acq., Rev. Rul. 76-411, 1976-2 C.B. 208. While the Commissioner believes that client groups may cancel their contracts at will, even an at-will relationship may constitute a valuable asset if it is reasonably likely to continue into the future. "[I]n valuing a contractual relationship, it is appropriate to determine the useful life of a contract by taking into account the likelihood of its renewal." Trigon, 215 F.Supp.2d at 715 (citing Union Bankers, 64 T.C. 807; Super Food Servs. v. United States, 416 F.2d 1236 (7th Cir.1969)); cf. Newark Morning Ledger, 507 U.S. at 550 n. 4, 113 S. Ct. 1670.

The more difficult question is whether the contracts are part of one or more "mass assets" or (synonymously) "indivisible assets." In his notice of deficiency, the Commissioner alleged that Capital's "specific individual customer contracts are components of intangible assets which constitute single indivisible assets, including customer-based intangibles." This theory relies on the hoary "mass asset rule," which limits a taxpayer's ability to deduct losses of some intangible assets that are treated as mere components of a larger, indivisible asset.

"The mass-asset rule prohibits the depreciation of certain customer-based intangibles because they constitute self-regenerating assets that may change but never waste." Newark Morning Ledger, 507 U.S. at 558, 113 S. Ct. 1670. While the Newark Morning Ledger Court discussed depreciation of the entire mass asset, the rule also prevents taxpayers from deducting the loss of individual components of the mass, as those deductions would for most practical purposes be equivalent to depreciation deductions. See Sunset Fuel Co. v. United States, 519 F.2d 781, 783-84 (9th Cir.1975).

In Newark Morning Ledger, the Supreme Court allowed the taxpayer to take depreciation deductions for its subscriber base, finding that these "paid subscribers" "constituted a finite set of subscriptions" and were not "composed of constantly fluctuating components." 507 U.S. at 567, 113 S. Ct. 1670. This distinguished them from a mass asset. The Court also cited Ithaca Industries, Inc. v. Comm'r, 97 T.C. 253, 1991 WL 151392 (1991) (Ithaca I), which denied a deduction for "assembled work force," finding that this force was a nondiminishing asset and that "new employees were trained in order to keep the `assembled work force' unchanged, and the cost of training was a deductible expense." Newark Morning Ledger, 507 U.S. at 560, 113 S. Ct. 1670 (citing Ithaca I, 97 T.C. 271).

In the wake of Newark Morning Ledger, the mass asset rule is on somewhat uncertain footing. Indeed, if the subscription lists in that case were not a mass asset, it is arguably difficult to see what would be. The leading post-Newark Morning Ledger mass asset case is Ithaca Industries, Inc. v. Comm'r, 17 F.3d 684 (4th Cir.1994) (Ithaca II), which affirmed the Tax Court's judgment in Ithaca I. The Fourth Circuit framed the issue as follows:

As the Court suggested in Newark Morning Ledger ... the distinguishing feature of a true mass asset is its ability to be regenerated without substantial effort on the part of its owner, that is, its ability to "self-regenerate." When an asset is maintained only by significant affirmative efforts to add new elements, these additions are most naturally understood as comprising something new and distinct from the original asset.

Ithaca II, 17 F.3d at 688. This test has been adopted by the Tax Court. See Hardware Plus, Inc. v. Comm'r, 67 T.C.M. 3045, 1994 WL 237350.

Although the question is not without its difficulties, it appears that Capital's contracts do not constitute part of a single indivisible asset. The Commissioner does not even contend that Capital's contract base is able to "self-regenerate." See Ithaca II, 17 F.3d at 688. Instead, Capital must make "significant affirmative efforts" to acquire new group contracts, id., although the costs of these efforts are generally deductible, see Ithaca I, 97 T.C. 271. Moreover, the insurance contracts seem to be distinct in fact: each contract has a unique economic value based on its claims history, rate structure, group size, and other characteristics. We thus are satisfied that these contracts are not a mass asset.3

It appears, thus far, that Capital has at least a theoretical right to take § 165 loss deductions when its group insurance clients cancel contracts that were in force on January 1, 1987. Furthermore, Capital's basis in each of these contracts—and thus the amount of each deduction—is equal to the fair market value of each contract as of January 1, 1987. That said, the calculation of this fair market value is difficult and hotly disputed, as is the question who bears the burden of proof.

The burden of proof in a Tax Court case is on the petitioner, but the Commissioner bears the burden of proof "in respect of any new matter." Tax Ct. R. 142(a)(1). The Tax Court placed the burden of proving the contracts' value on Capital. See 122 T.C. 246 n. 11.

Capital, however, contends that the Tax Court should have shifted the burden of proof to the Commissioner, because "the Notice of Deficiency did not raise the issue of Capital's valuation," or in the alternative because the notice "was arbitrary for finding no value."

The first of these contentions is based on the fact that the notice denied that Capital had established "any" loss. Capital claims that the dispute over its specific valuation is therefore "new matter" within the meaning of Rule 142. The Tax Court found that the notice's "broad language relating to whether Petitioner sustained `any loss' ... includes the factual valuation issue." 122 T.C. 246 n. 11. We agree with the Tax Court: the notice of deficiency challenging Capital's entire deduction necessarily raised the valuation issue at the heart of that deduction.

Capital's more compelling contention is that the Commissioner's initial notice of deficiency denying any deduction was per se unreasonable. In deficiency cases brought in Tax Court, the petitioner may shift the burden of proof to the Commissioner if it can prove that the Commissioner's assessment is "arbitrary and excessive." Helvering v. Taylor, 293 U.S. 507, 515, 55 S. Ct. 287, 79 L. Ed. 623 (1935). Even the Commissioner's experts agree that these contracts had some value. Therefore, Capital argues, it has demonstrated that the Commissioner's zero-deduction position is unreasonable, and so the burden shifts to the Commissioner to prove the correct amount of the deductions.

Taylor, however, concerned the apportionment of income, not the calculation of deductions. Indeed, the Taylor Court stated in dictum that "the burden is upon the taxpayer to establish the amount of a deduction claimed," even where the Commissioner's position is unreasonable. Id. at 514, 55 S. Ct. 287.

We have applied Taylor to deduction claims in circumstances closely analogous to this case. In R.M. Smith, Inc. v. Comm'r, 591 F.2d 248 (3d Cir.1979), the Commissioner disputed the amount that the taxpayer could deduct for amortization of certain patents. The taxpayer's basis for deductions depended upon the fair market value of the patents at the time they were acquired. The taxpayer and the Commissioner put forward widely varying calculations of this fair market value; the Tax Court, rather than choosing one or the other, split the difference and entered a judgment based on its own valuation. The taxpayer appealed, arguing that (1) once it had proven that the Commissioner's valuation was "unduly pessimistic," the burden shifted to the Commissioner, and (2) because the Tax Court rejected the Commissioner's valuation, it was required to accept Smith's valuation. We agreed with the first of these contentions:

Smith states that it had the burden of proving that the Commissioner's determination of the gross values, which initially amounted to $10,000, was arbitrary. Having established that the calculation was in error, Smith was not required to prove the correct figure. The Commissioner had the burden of establishing the proper valuation and thus the actual tax owed. This much is an accurate statement of the law.

Id. at 251 (citing Taylor, 293 U.S. 507, 55 S. Ct. 287, 79 L. Ed. 623; Fed. Nat'l Bank v. Comm'r, 180 F.2d 494, 497 (10th Cir.1950)).

But we rejected Smith's second contention that the Tax Court was bound to award the deduction requested by Smith merely because the Commissioner's valuation was unreasonable. We continued:

Where Smith's argument fails is in suggesting that the refusal of the tax court to accept the Commissioner's evidence requires this court to reverse the trial judge with instructions to expunge the deficiencies. The teaching of Helvering v. Taylor ... and Federal National Bank v. Commissioner of Internal Revenue ... is that the appropriate remedy in the absence of evidence of proper valuation is a remand to allow for additional evidence to be presented. In this case, however, sufficient evidence was introduced to allow the tax court to reach a reasonable conclusion. The court is not limited to simply choosing one of the two values proffered. It is appropriate for it to evaluate all of the evidence and to make an independent determination that does not necessarily accept the valuation of either party.

Id. Indeed, in Taylor itself, the Supreme Court did not reject the Commissioner's arbitrary claimed deficiency; instead, it affirmed the decision of the Court of Appeals remanding the case so that the Board of Tax Appeals might "hear[] evidence to show whether a fair apportionment might be made and, if so, the correct amount of the tax." 293 U.S. at 516, 55 S. Ct. 287. In sum, the taxpayer bears the burden of proof on valuation, but where the Commissioner's alternative proposed valuation is unreasonable, the Tax Court must make a fair apportionment, and is not confined to choosing between the two proffered valuations.

Nonetheless, as Taylor recognized, it is not always possible for the Tax Court to make a "fair apportionment" of a taxpayer's obligations and basis. 293 U.S. at 516, 55 S. Ct. 287. Capital argues that its burden is only to prove that its aggregate deductions are correct, even if it cannot apportion them precisely. See S. Pac. Trans. Co. v. Comm'r, 75 T.C. 497, 1980 WL 4591 (1980); DiLeonardo v. Comm'r, 79 T.C.M. 1820 (T.C.2000). The Commissioner disagrees, arguing that Capital must prove the value of each contract, and may not take an approximate deduction not based in the value of the contracts actually lost.

Newark Morning Ledger, which presents significant factual analogies to this case, set a high bar for the taxpayer seeking to prove deductions. The Court held that the taxpayer bears the burden of proving "that a particular asset can be valued and that it has a limited useful life," and noted that "that burden will often prove too great to bear." 507 U.S. at 566, 113 S. Ct. 1670.

Most mass asset cases before and after Newark Morning Ledger make similar statements. These cases put the burden on the taxpayer of "establish[ing] reasonably accurately a basis in the particular account on which the loss is claimed." Sunset Fuel, 519 F.2d at 783; see also 7 Mertens, Law of Federal Income Taxation § 28.15, at 49-50 (rev'd ed.2001). Some of these cases go even further, echoing Newark Morning Ledger's warning that the burden will often be too great for the taxpayer to bear. See, e.g., Globe Life & Accident Ins. Co. v. United States, 54 Fed. Cl. 132, 136 (2002).

There is some tension between the rule of Newark Morning Ledger, which places on the taxpayer a "heavy burden" of proving that its intangible assets may be valued separately, and the rule of Taylor and R.M. Smith, which allows the Tax Court to determine a fair value when neither the Commissioner's nor the taxpayer's valuation is completely convincing. We think, however, that these cases can be reconciled in a straightforward manner.

Newark Morning Ledger stands for the proposition that the taxpayer always bears the burden of proving that his lost intangible assets are susceptible of separate valuation. A taxpayer who cannot carry that burden possesses a mass asset, and may not depreciate it or deduct losses of components of that asset. The Commissioner may always put the taxpayer to his proofs in such a case, and the Commissioner's litigation position rejecting the entire claimed deduction will not necessarily be unreasonable.

But this heavy burden applies only to the taxpayer's obligation to prove that his intangible assets may be valued separately and with reasonable precision. What we derive from the foregoing is that, if the taxpayer can satisfy that burden, the process of proof changes. Once a court is satisfied that the intangible assets may be valued separately, its obligation is to find the correct value. The taxpayer and the Commissioner may submit their own proposed valuation, and dispute over the merits of each side's claims. Where the Commissioner refuses to submit any valuation, or where his valuation is arbitrary, the court will essentially be forced to start from the taxpayer's valuation.

The court may accept this valuation if it is reasonable, or it may modify it to take into account objections raised by the Commissioner or by the court sua sponte. But it will not, in our view, be reasonable for a court to reject the taxpayer's valuation out of hand simply because the Commissioner has identified minor flaws in the valuation. Once it is established that the assets have a reasonably ascertainable value, the court is obligated to seek the correct value of the contracts not, upon catching the taxpayer in an error, to deny any deduction automatically. See R.M. Smith, 591 F.2d at 251.

Similarly, the dispute over whether Capital must prove only its aggregate 1994 deduction, or the individual value of each contract lost in that year, is a chimera. Under Newark Morning Ledger, Capital must of course prove that each of the 376 lost contracts has an individual value that exists separately from that of the other contracts. But the court need not find each individual valuation convincing in all respects in order to accept an aggregate deduction. A court might find that ten individual contracts each have a separate value, while being unable to put a precise dollar value on each one. In such a case, if the court can easily calculate the aggregate value of the ten contracts, while remaining uncertain about the individual values, the taxpayer has satisfied its burden under Newark Morning Ledger and may prove his aggregate deduction under the logic of South Pacific Transportation and DiLeonardo.

The Tax Court's "finding of fact with respect to valuation is to be reviewed under the clearly erroneous standard." R.M. Smith, 591 F.2d at 251. Here, the Tax Court concluded that Capital had failed to meet its burden of proving the 1987 fair market value of its lost contracts. Its basis for this conclusion was a litany of perceived flaws in Capital's valuation. Most importantly, the Tax Court found that Capital's "reinsurance model" for valuing the lost contracts was flawed, 122 T.C. 249-51; that the valuation did not completely take into account individual characteristics of the group contracts, id. at 251-55; and that the valuation was based on defective assumptions regarding the expected life of the contracts, id. at 255-57. We address each of these findings in turn. On appeal, the Commissioner has added numerous instances of perceived error in Capital's valuation; we address those claims in the course of our consideration of the Tax Court's conclusions.

The Tax Court's most significant criticism of Capital's valuation concerned the "reinsurance model" used by Daniel McCarthy, Capital's actuarial valuation expert, see supra n. 2.

McCarthy valued the contracts at issue at their "highest and best use." He concluded that this use could be determined using a "reinsurance model," whereby he asked how much the 376 contracts cancelled in 1994 would have been worth if they had been sold together in a reinsurance transaction in 1987. McCarthy argued that this is the highest and best use of the contracts, and that they would be worth more sold together in a reinsurance transaction than they would be if they were sold separately. In fact, it seems unlikely that any such contracts could be sold separately. See Trigon, 215 F.Supp.2d at 706 ("It is an undisputed fact that there are no known sales of single group health insurance contracts between insurance carriers either before January 1, 1987, or after."). Blocks of contracts can, however, be sold, and McCarthy represented that the 376 contracts lost in 1994 would have constituted a "credible" block that could have been sold between insurers. See 122 T.C. 250.

The Trigon court reasoned that these contracts are properly valued under a willing-buyer model that assumes a buyer with facilities comparable to those of Capital:

[Group] contracts must be valued on the theory that they would be sold to a hypothetical willing buyer having facilities comparable to those of the seller. Attempting to value the contracts on a stand-alone basis (which the government appears to advocate), rather than as part of a going concern, results in an improper determination of "liquidation value," rather than fair market value.

215 F.Supp.2d at 708-09 (citation omitted). We find this analysis persuasive.

We reject the government's argument that a one-at-a-time sale model is required. The government cites cases holding that minority shares of stock must be valued according to their own value, without taking into account a control premium that might inhere in a larger block of stock. See Ahmanson Found. v. United States, 674 F.2d 761, 772 (9th Cir.1981); Rev. Rul. 93-12, 1993-1 C.B. 202. But such cases are inapposite: the control premium for a majority stock holding has a separate value over and above the value of each individual share, while McCarthy's use of the reinsurance model is designed not to capture additional value but to account for the transaction costs involved in selling a single contract.4 Capital need not value its contracts only at their liquidation value, rather it may use the reinsurance model to determine its basis.

The Tax Court nonetheless took the reinsurance model as evidence that Capital's contracts could not be valued individually. It found that, at most, McCarthy had proven the value of the 376-contract block sold by Capital, but it had not proven the value of each individual contract:

[A]s it must, petitioner does not claim a single loss deduction in 1994 upon the termination of the 376 group contracts. Rather, petitioner claims 376 separate loss deductions relating to the termination of each of the 376 separate group contracts. What is required to support petitioner's claimed loss deductions under section 165 are valuations of the group contracts that reflect a value for each contract as a separate and discrete contract.... [A]ll petitioner has done is establish that the group contracts are capable of being valued in blocks. Petitioner has not, however, established that the group contracts are capable of being valued separately and independently as individual assets.

122 T.C. 250-51; see also Trigon, 215 F.Supp.2d at 709 ("[T]he issue is not whether the highest and best use of Trigon's contracts is as part of an ongoing health insurance company.... The issue, instead, is whether specific contracts can be valued separately from the block of contracts to which they belong.").

Capital argues that the contracts can be valued separately, and that McCarthy did in fact value each contract separately. McCarthy testified to this effect, noting that his valuation methodology in this case was consistent with his practice in appraising insurance contracts when advising insurers that are demutualizing:

It was consistent, in that it took into account the characteristics of each contract being valued. It was consistent, in that it took into account as the standard of that to be discounted of the emerging stream of expected statutory earnings, and it was consistent, in that they were discounted to present value.

He referred to this as a "seriatim or one-at-a-time valuation." He readily admitted, however, that he calculated the contracts' value based on an assumption that they would be sold in batches.

We think that the Tax Court, and the Commissioner, misunderstood the requirements of separate valuation. As noted above, Newark Morning Ledger, 507 U.S. at 566, 113 S. Ct. 1670, requires that a taxpayer wishing to deduct his losses of intangible assets must show that those assets are susceptible of separate valuation. In many cases, this will be impossible, simply because the taxpayer really possesses a single indivisible asset whose whole is incommensurable with the sum of its parts, a single mass "composed of constantly fluctuating components." Id. at 567, 113 S. Ct. 1670. Thus, for instance, a company may not depreciate its "assembled work force," because new employees are constantly being trained to replace old ones, and because there is no meaningful way to assign distinct values to each member of this workforce. Id. at 560, 113 S. Ct. 1670. The value inheres in the "assembly" of the workforce, not in any one individual.

Insurance contracts are different. They are valued all the time; indeed, Daniel McCarthy, Capital's expert, has spent much of his career valuing health and life insurance contracts in order to advise insurers and regulators on the fairness of demutualization transactions. While the Tax Court and the Commissioner have numerous quibbles with Capital's valuation, they do not persuade us that these contracts do not each have an individual value. As Capital succinctly puts it, "the Tax Court erred because it confused (1) the question of whether an intangible has a value and useful life separate from goodwill..., with (2) the question of what the asset's value is."

The Commissioner cites several pre-Newark Morning News cases for the proposition that taxpayers may not use average values to compute the value of specific accounts. Sunset Fuel, 519 F.2d at 785-86; Skilken v. Comm'r, 420 F.2d 266, 270 (6th Cir.1969). But the averaging procedures in those cases were far cruder than McCarthy's sophisticated statistical methods here. McCarthy represented that the 376 contracts lost in 1994 constituted a "fully credible" block of contracts, such that a willing-buyer reinsurer would expect high— and low-value contracts to cancel each other out, and would therefore purchase the community-rated contracts based on average rather than individual experience. Experience-rated contracts were, at all points, valued individually.

The evidence is clear that McCarthy's voluminous, thorough, and professional valuation was meant to determine a value for each individual insurance contract. As part of that individual valuation, McCarthy used various averaging procedures, sometimes to check his work, but sometimes as part of his initial calculations. The undisputed evidence appears to be that such averaging procedures were consistent with industry standards for valuing group insurance contracts for the purposes of reinsurance or demutualization. McCarthy's use of industry-standard statistical methods does not render his appraisal invalid, or support the Tax Court's conclusion that Capital's contracts could not be valued individually. We thus hold that that conclusion was clearly erroneous.

The Tax Court also rejected McCarthy's reinsurance model on the grounds that McCarthy made only "some type of vague expense adjustment" to account for intangibles such as goodwill that were associated with the 376 terminated contracts. 122 T.C. 250-51. A larger adjustment for the intangibles, which the Commissioner believes is justified, would lead to a smaller tax deduction.

McCarthy subtracted some $300 million from his total valuation of all of Capital's contracts, to account for the value added by Capital's name, reputation, and other goodwill factors, as well as by its workforce and provider network. These factors made up a part of the value of Capital's contracts, but were not lost when those contracts were lost; therefore, Capital did not—and could not—claim them as part of its deduction. The dispute here is over the method of calculating this goodwill adjustment. McCarthy used a rental charge, whereby he valued these intangibles based on what it would cost Capital to rent them in a market transaction. The Commissioner argued, and the Tax Court agreed, that this was improper. Instead, the Tax Court found that McCarthy should have deducted a "capital charge" from the value of the contracts, based on a valuation of the intangible factors that takes into account the market rate of return on those factors. Although the Commissioner has not attempted to calculate what such a charge would look like, we assume that it would lead to a greater offset for these intangibles, and so to a smaller tax deduction.

Capital argues, however, that McCarthy's method, which used a rental charge rather than a capital charge, was proper and indeed standard. The Tax Court found that McCarthy's explanation for not taking a capital charge for the related intangibles was "not credible." Id. at 251. Capital claims that this contradicted the "undisputed testimony of all the experts," which was that rental charges are normally used in insurance valuation, and that McCarthy's use of them was proper. The Commissioner responds that McCarthy's charge for the related intangibles was based on their cost to Capital rather than their market value, and that this method of deducting the other intangibles overstated the value of the lost contracts.

Capital's characterization of the record appears to be mistaken. The Commissioner's witnesses did not concede that McCarthy's approach was correct, although neither did they claim that it was professionally untenable. They did argue for an alternative method, which presumably would have given a different, and greater, value to Capital's goodwill.

Given the dispute in the record between well-qualified experts, and the Tax Court's greater familiarity with the issue, we cannot conclude that the Tax Court's finding here was clearly erroneous. Indeed, this finding seems conceptually correct—Capital's goodwill factors should be subtracted at their value, not their cost—although Capital argues that McCarthy's rental charge was meant to estimate value and not cost. We thus accept the Tax Court's conclusion that McCarthy should have used a capital charge, rather than a rental charge, to extract goodwill from his valuation of the contracts.

But, as explained above, see supra Part IV.B, Capital does not lose its entire deduction merely because the Commissioner has found some flaws in its method. On the remand that our other holdings require, the Commissioner will have the opportunity to explain what McCarthy should have done differently in this regard, relying on specific calculations of cash flows rather than on generic names like "capital charge" versus "rental charge." The Commissioner will also be able to propose an alternate valuation for the $300 million goodwill adjustment. Capital, meanwhile, will have another opportunity to demonstrate to what extent McCarthy's method captures the factors raised by the Commissioner.

Ultimately, the Tax Court must determine what goodwill adjustment is appropriate, using either McCarthy's rental charge, a capital charge proposed by the Commissioner on remand, or some other adjustment taking into account the arguments of both sides. In sum, the mere fact that McCarthy's charge is flawed does not mean that the Tax Court may reject Capital's entire valuation.

The Tax Court next found that Capital's "expert utilized incomplete information and made erroneous assumptions relating to the characteristics of the group contracts that alone would support disallowance of the $4 million in loss deductions claimed." 122 T.C. 251. The Tax Court identified several instances of what it considered incomplete data or erroneous assumptions, and found that these errors made McCarthy's valuation so uncertain as to be almost meaningless.

Most importantly, the Tax Court found that McCarthy "[i]gnored or did not consider historical premium payment and claim patterns and renewal expectations relating to each contract." 122 T.C. 251-52. The Tax Court also noted that, for many of the contracts that he valued, McCarthy used average premium or claims data, because individual contract data was lacking. Id. at 252-53.

Capital argues that the Tax Court was in error, because the experience-rated contracts were appraised based on "premium rates in effect on January 1, 1987, which reflect each of these factors on a contract-specific basis." Capital expert Constance Foster, an insurance attorney and former Insurance Commissioner of Pennsylvania, testified that "The demographics or any information about the customer is embedded in the rate. All we would do in renewing is look at their past claims experience and how many people are represented to project new rates." Put differently, the experience rating process was intended to capture each group's claims and payment experience in calculating each year's premium rate. Thus, McCarthy was able to value the experience-rated contracts using only rate information, because the rate information captured the information that the Tax Court found missing. Capital also explains that some 24 of the experience-rated contracts, which together accounted for approximately $2.5 million of its $4 million claimed deduction, were valued taking into account all of those contract-specific factors, without any averaging or missing data regarding premiums or claims.

As an example of McCarthy's failure to accurately appraise the value of experience-rated contracts, the Tax Court cited the case of Pennsylvania Farmer's Union. This client group maintained three experience-rated contracts that had a cumulative deficit of some $700,000 in January 1988. By 1994, the deficit was $4 million, and the contract was cancelled the following year, leaving Capital to absorb the deficit. 122 T.C. 255. The Tax Court noted that, despite these deficits, McCarthy assigned the three Farmer's Union contracts "a total positive value of $479,000, or nearly 20 percent of the total value attributed to all of petitioner's experience-rated group contracts that were terminated in 1994." Id.

The Tax Court's reasoning is flawed because it assumes that the contract was not of positive value in 1987 merely because the contract ultimately caused Capital a loss. As Capital persuasively observes, it would not have renewed experience-rated contracts that it expected to result in a loss, so it can reasonably be assumed that in 1987 its contracts had a positive expected value. Indeed, one of Capital's experts testified that deficit accounts have "a bit of more value" "as long as they stay with Capital Blue Cross," because such accounts "produce[] more margin" in that Capital will raise premium rates going forward for contracts with high claims rates. Testimony also indicated that such deficit contracts did stay with Capital Blue Cross, or at least that their lapse rates were not materially greater than those of other contracts.

As it turned out, Capital continued to lose money on the Farmer's Union contract, and the group ultimately cancelled the contract rather than paying a significantly increased premium, leaving Capital with a large accumulated deficit. Capital's Chief Financial Officer, Robert Markel, testified that Farmer's Union had a deficit that was not "unusually large" as of 1987, which grew larger in later years. He also testified that Farmer's Union breached its contract by transferring low-risk employees to another carrier, and that Capital cancelled the contract in 1994 upon learning of the breach. Capital submits that, if this breach had not occurred, it would have been able to recoup its losses from the contract and make it profitable. Capital's position is that Farmer's Union's ballooning deficit and contract breach were not foreseeable in 1987; therefore, a reasonable buyer would have assigned the contract a relatively large positive value.

We believe that McCarthy's valuation was reasonable. First, some 60% of Capital's claimed deductions come from experience-rated contracts in which averaging was not used and the data is complete. His assumptions about experience-rated contracts, including those with a deficit in 1987, were economically sensible; the fact that the Farmer's Union contract turned out to be disastrous does not prove that it had zero or negative value as of January 1, 1987.

The Commissioner also seeks to defend the Tax Court's decision about the characteristics of group contracts by pointing to the community-rated contracts, for which significant averaging was used. He notes, as did the Tax Court, that a small change in the expected claims ratio could have an enormous effect on the expected value of the contract: "use of a claims ratio just 1 percent higher than the aggregate average claims ratio used by petitioner's expert for community-rated group contracts would reduce petitioner's projected profit relating to the contracts by more than half." 122 T.C. 254. But the Commissioner has not demonstrated that there is anything wrong with using averaging, and the evidence indicates that an actual willing-buyer reinsurer would use an averaging method essentially identical to McCarthy's.

The Tax Court is, of course, correct that averaging is sensitive to initial assumptions: here, for instance, the valuation of the community-rated contracts depends on the average claims ratio; the appropriate discount rate will also have a significant effect. But there is little in the record before us to suggest that McCarthy should have used different assumptions. In view of this fact, we must remand to the Tax Court to allow it to consider McCarthy's procedure in more detail.

On remand, the Commissioner may explain his objections to specific assumptions in McCarthy's valuation, as regards both individual experience-rated contracts and the collective assumptions about community-rated contracts. If the Tax Court finds that these assumptions were incorrect, it may find more appropriate figures and use them to calculate a more appropriate valuation of Capital's contracts. But if McCarthy's assumptions were correct, or were those that a reasonable buyer would make, then the fact that his calculations were sensitive to his assumptions does not render his valuation incorrect.

A central part of McCarthy's valuation was his "lifing analysis," that is, the method by which he estimated the expected life of each contract as of 1987. McCarthy used historical lapse rates to determine the probability that each group's contract would lapse in any given year. These lapse rates were used to compute the expected life of each contract, which was essential to calculating its fair market value.5 McCarthy used historical data from Capital's 1982-1986 experience that "indicated that each group contract had a 2.2-percent to 7.5-percent probability of lapsing from year to year, depending on factors such as group size and duration of the contract." 122 T.C. 255. The Tax Court had two objections to McCarthy's lifing analysis: first, that McCarthy did not take into account the uncertainty in the insurance market in 1987, and second, that he did not take into account certain "human elements" that influence lapse rates.6

First, the Tax Court found that the lapse rates did "not account for foreseeable (as of January 1, 1987) and significant changes in the health insurance marketplace." 122 T.C. 255-56. More specifically, the Court found that McCarthy did not consider the impact that increased competition from HMOs and other new insurance products would have on Capital's lapse rates. In 1987, the argument goes, a willing buyer could have predicted that increased competition would lead to greater lapse rates, and thus would have valued Capital's contracts at a rate lower than McCarthy ascribed to them using historic lapse rates.

Capital submits that this line of reasoning reflects a misunderstanding of the evidence. Capital's CEO testified that, in the 1980s, Capital was aware of the competition from HMOs, but expected that this competition would not significantly affect its market share or lapse rates because central Pennsylvania, where Capital operates, has a traditional market with relatively few hospital choices and a strong organized-labor presence. McCarthy testified that he took the competitive situation into account, but determined that most of the "competitive factors" that led to lapses had already taken effect in the 1982-1986 period that he used to determine lapse rates.

The Commissioner's experts reasoned that "[i]n the presence of such significant market changes, the assumption that future lapse rates would be consistent with past lapse rates is, at best, problematic," and that the lapse rates were "speculative in the extreme, given what was going on in the group health insurance market at the time." We are unconvinced. First of all, McCarthy, unlike the Commissioner's experts, seems to have spoken to Capital management and considered circumstances unique to Capital's central Pennsylvania market, while the Commissioner's experts considered only the national health insurance market. Capital presented evidence that its market was (for reasons suggested above) uniquely resistant to the competitive pressures introduced by HMOs and PPOs; therefore, McCarthy's calculations based on Capital's own past data may well have been more accurate than the Commissioner's projections based on national trends.

Furthermore, there does not seem to be any evidence that McCarthy's lapse rates were incorrect. Instead, Capital's evidence tends to show that subsequent experience proved Capital's projections correct: it has not lost significant market share to HMOs, and its historic lapse rates from 1987 were very close to those predicted by McCarthy. McCarthy's table of historically derived lapse rate assumptions—ranging from a 2.2% lapse rate for groups of 10-24 members whose contracts had been in effect for over ten years to a 7.5% rate for groups of 1-9 members whose contracts had been in force for under one year — squares relatively well with Capital's actual 1987-1994 experience. McCarthy testified that the experienced lapse rate by number of contracts was 5.6%, while the lapse weighted rate by total premiums lost was just under 3%. As McCarthy explained, the former number corresponds reasonably well to his predicted lapse rates for small groups, which were the most numerous, while the latter corresponds quite well to his predicted rates for large groups, which made up the bulk of Capital's premiums. Of course, it is theoretically possible that McCarthy's predicted lapse rates were speculative, but nonetheless turned out to be correct. But the general accuracy of his predictions is certainly strong evidence that they had a foundation in reasonable analysis rather than speculation.

The Commissioner denies that McCarthy's lapse rate predictions were accurate, although he has not pointed to any statistical evidence to refute the numbers we have cited above. Instead, the Commissioner cites a Capital marketing plan from 1993, which states that "[s]ignificant losses from existing accounts are being incurred from HMO's," and that the Berkshire Health Plan PPO had "targeted Blue Cross and Blue Shield customers and [had] been successful in enrolling a significant number of accounts through selective underwriting."

This document certainly supports the Commissioner's thesis that Capital faced competition from HMOs. But McCarthy's testimony was that this competition had already developed by the 1982-1986 period that he used to estimate post-1987 lapse rates, and that his use of historical rates therefore accurately captured Capital's 1987 expectation of future rates. He testified that, based on discussions with Capital executives, he concluded that "the phenomena [of increased competition, including from HMOs and PPOs] had really already developed in the period of time I used for purposes of the lapse study.... And so I felt, after listening to them, that based on the situation in 1986, it was not necessary to modify those experience rates that i[sic] had derived for purposes of projecting in the future."

Without any contradictory evidence, we have no choice but to accept McCarthy's representations that his predicted lapse rates turned out to be accurate. Moreover, his procedure—relying on recent historical rates that he concluded incorporated the developing changes in the insurance industry, and discussing his predictions with Capital management to get a sense of their predictions as of 1987—does not strike us as "speculative in the extreme." The Tax Court appears to have ignored Capital's evidence that McCarthy's lapse rates were accurate, and to have unduly credited the Commissioner's experts' conclusory assertions that those rates were speculative. We thus reject as clearly erroneous the Tax Court's conclusion that "petitioner's expert largely ignored the industry changes of which petitioner's management, as of January 1, 1987, was aware." 122 T.C. 256-57.

Tax Court also rejected McCarthy's lifing analysis because it found that McCarthy did not consider various "human elements" that would influence lapse rates, viz., various subjective factors that might make customers cancel their at-will contracts. It held that "These human elements associated with petitioner's group contracts created a significant element of unpredictability with regard to the useful life of petitioner's group contracts." 122 T.C. 257.

The Tax Court's conclusion here reflects a fundamental misunderstanding of McCarthy's method, if not of the nature of the insurance industry and actuarial methods. McCarthy took into account the human factors in the way that all actuaries do: actuarially. He divided the contracts into groups based on distinctions that he found relevant, computed average lapse rates, and used them to project future lapse rates. Capital quite wittily cites Ehrhart v. Comm'r, 57 T.C. 872, 873, 1972 WL 2416 (1972) ("Actuaries are highly skilled mathematicians who deal with various contingencies affecting human life."), for the proposition that an actuary of McCarthy's experience is well equipped to deal mathematically with the human factors affecting the lapse rates of insurance contracts.

Furthermore, the Tax Court did not identify any "human factors" that McCarthy's valuation failed to take into account. The Tax Court's reliance on Ithaca II, supra, 17 F.3d at 689-90, and Globe Life & Accident, supra, 54 Fed. Cl. 132, is misplaced. Those cases concerned workforces, which are much harder to value than insurance contracts; the valuations involved there were far less careful and thorough than McCarthy's valuation here; and those cases concerned amortization (where precise lapse rates are essential) rather than deductions for the direct loss of contracts.

Because it ignored undisputed evidence and misunderstood the nature of McCarthy's calculations, the Tax Court's rejection of Capital's lifing analysis on the theory that McCarthy failed to take into account any subjective "human factors" was clearly erroneous and must be set aside.

In his appellate briefs, the Commissioner builds on the Tax Court's findings by arguing that McCarthy's lapse rate assumptions were flawed in other respects. Specifically, McCarthy only used average rates for contracts of a given size and age, and did not calculate different rates for different kinds of coverage or contract, different premium payment histories, changing sizes, or financial condition of the client group. Capital responds that McCarthy complied with actuarial principles in coming to his conclusions, and that the Commissioner has not shown that McCarthy's valuations would be different if he took into account the more specific factors that the Commissioner urges.

We agree with Capital. McCarthy's efforts were thorough, and it appears to be undisputed that he followed actuarial standards. The Commissioner has identified some factors that he did not consider, but this alone does not seem to be a reason to reject McCarthy's lapse rate calculations. As the Tax Court has previously stated, "lapse rates may be determined from a statistical analysis of actual past experience of policies in force at specified intervals of time or from an informed judgment of a person who has had experience in the field." Union Bankers, 64 T.C. 816.

Simply put, it would be impossible for McCarthy to take into account every factor that might distinguish one contract from another. McCarthy did not classify contracts based on what percentage of individuals in each group was left-handed, but the Commissioner would not be heard to argue that this was a flaw in his methodology. The Commissioner cannot invalidate McCarthy's methodology simply by pointing to factors that McCarthy neglected; instead, he must also make a reasonable case that such a factor would have changed his conclusions. The Commissioner has not even attempted to do so here, and we see no reason to reject McCarthy's lifing analysis.

In his appellate brief, the Commissioner has not confined himself to defending the Tax Court's opinion on its own terms. He has also put forward several other purported grounds for affirmance. The Commissioner now objects to the completeness and accuracy of Capital's records, arguing that coding errors and missing data render many of Capital's conclusions suspect. He also argues that McCarthy drew improper inferences from the aggregate value of Capital's 23,526 contracts in 1987 to the value of the 376 contracts cancelled in 1994, without first demonstrating that those 376 contracts were a representative sample of the whole.

Without the benefit of explicit factual findings by the Tax Court on these issues, we will not undertake to decide them. Instead, we will allow the Tax Court to consider these arguments in the first instance. This consideration will involve both a determination whether the Commissioner is correct about the alleged flaws in Capital's data and methodology and a decision about the extent to which those flaws invalidate McCarthy's ultimate valuation.

Two themes emerge from the above discussion. First and foremost, we have rejected as clearly erroneous the Tax Court's ultimate conclusion that Capital "has not ... established that the group contracts are capable of being valued separately and independently as individual assets." T.C. 251. We find that McCarthy's model, including his use of some averaging assumptions, established an individual value for each contract with sufficient specificity to carry Capital's burden under Newark Morning Ledger. See supra Part IV.B.

Second, we have rejected as clearly erroneous some, but not all, of the Tax Court's specific findings of flaws in Capital's valuation. Because of the centrality of these findings to the decision we are constrained to reverse and remand. Because it is clear that Capital had some basis in the contracts, we do not think that even the Tax Court's valid objections prevent Capital from taking a deduction. On remand, the Commissioner will have the opportunity to quantify his objections to Capital's valuation, and the Tax Court will be able to decide the proper valuation. Thus, for instance, the Commissioner may dispute McCarthy's goodwill adjustment by proposing his own capital charge, see supra Part V.A.3, and the Tax Court may determine what the appropriate goodwill adjustment should be. Similarly, the Commissioner may explain which of McCarthy's initial assumptions—about contract claims rates, discount rates, etc.—were erroneous, see supra Part V.B, and the Tax Court may adjust McCarthy's valuation if it finds that his assumptions need to be changed.

That said, we expect that the Commissioner will not continue to rely solely on experts who testify that the lost contracts are impossible to value: without a competing valuation argument, it would seem that the Tax Court will have little choice but to grant Capital its claimed deduction. As we have stressed above, see supra Part IV.B, once Capital has carried its burden of showing that the contracts may be valued individually—as we believe it has—the Tax Court's role is to find the correct valuation. Because there is no real dispute that the contracts had value in 1987, and because we find that they may be valued individually, a valuation of zero is unlikely to be the correct result.

Because we find that Capital's lost insurance contracts are susceptible of individual valuation as of January 1, 1987, the Tax Court's conclusion that Capital is not entitled to a deduction for the loss of those contracts must be set aside. We reverse and remand for a determination of Capital's basis in those contracts, informed by the record and any further submissions from the parties that the Tax Court may consider appropriate.